Cutting Job Change Time 40-60% in a Container Glass Plant

It's 2am on a Wednesday. Section 4 has just come back up after a blank-and-mould change on a six-section IS machine. The moulds are hot, the machine is running, but gob weight CV is sitting at 0.9% and you've already had two leaners come off the annealing lehr. Your hot-end superintendent is chasing plunger drift on three sections simultaneously. The job change clocked 4 hours 20 minutes. It was supposed to be 2 hours 45.

That 1 hour 35 minutes of unplanned extension is where most container glass plants live. It's recoverable. That's the point.

Most plants are measuring the wrong thing

When I audited a two-furnace, six-line plant in regional Queensland two years ago, the plant manager could quote average job change time to the minute. What he couldn't tell me was how much of that time was actual mechanical change versus preparation work that should have been done before the IS machine stopped. That distinction is the core of SMED: separating internal activity, which requires a stopped machine, from external activity, which doesn't.

In container glass, this ratio sits badly. On most lines I walk, 35-45% of what happens inside the job change window could have been done beforehand. Mould pre-heating. Receiving inspection on incoming equipment. Confirming the recipe file is correct on the section controller. None of that requires downtime. All of it routinely happens after the IS machine stops anyway.

One number tells me whether a plant has this discipline: what percentage of job changes start with all moulds already at the preheat target? On the Queensland plant, it was 12%. After 90 days of pre-change preparation work, it was 84%. Average job change time dropped from 4 hours 8 minutes to 2 hours 31 minutes. That's -39% without touching the actual mechanical change procedure.

Where the time actually goes

Pull the timestamps from your last 10 job changes. Not the shift log summary. The actual timestamped activity record. On lines without structured recording, that data largely doesn't exist, and that already tells you something.

When I've done this exercise with operators, the breakdown tends to look like this:

• Pre-change preparation sitting inside the downtime window instead of before it: 45-70 minutes
• IS machine mechanical change: 60-90 minutes
• Post-change tuning to stabilise gob weight CV at ≤0.4% and section timing within 10ms: 40-80 minutes
• Defect-driven interruptions from settle waves or baffle marks: 20-60 minutes

The tuning tail surprises most managers. Getting gob weight CV back to ≤0.4% after a wide-mouth press-and-blow change can take as long as the mechanical change itself when recipe lock isn't clean. Operators tuning section by section, each adjustment drifting the next. The hot-end superintendent owns recipe lock. Operators don't change set points without sign-off. That single rule is worth 25-30 minutes on a typical job.

Baffle marks after a job change usually point at baffle alignment drift from the previous run, not the new tooling. Check the centring tool reading before you pull the old moulds. You'll save 40 minutes of chasing a defect that had nothing to do with the change itself.

Why shift handover kills your preparation window

The job change doesn't start when the IS machine stops. It starts 72 hours earlier. Most of the time you lose is in the preparation window, not the mechanical change.

SMED theory is clean. Application in container glass is messy because the activity spans multiple shifts, multiple roles, and a preparation window that begins at T-72h. Without a stage-gated structure, preparation discipline collapses at shift handover.

The 0600 handover misses pre-change preparation status 70% of the time on most lines I audit. Not because operators don't care, but because there's no structured field to hand it over in. "Jobs are running fine" doesn't tell the incoming superintendent that section 6 blanks haven't been pre-heated yet, or that the receiving inspection on new moulds flagged a seam-to-seam alignment issue at T-48h.

The 9-stage Job Change Lifecycle is built around this problem. It maps the full window from T-72h to T+45 min post-change stabilisation, assigns ownership at each stage, and gives shift handover a structured checkpoint rather than a verbal summary. The stage gates don't slow the change down. They compress the internal window by front-loading everything that doesn't require a stopped machine.

What a 40-60% reduction actually takes

A 40% reduction in job change time is achievable in 90 days on most container glass lines. The 60% figure takes 6-12 months and requires cultural work on top of procedural work.

The 90-day gains come from pre-change preparation discipline, recipe lock enforcement, and mould preheat verification as a hard gate before IS machine stop. The target is simple: all moulds confirmed at 480°C ±10°C before the stop order goes out. No exceptions. These don't require capital. They require a superintendent who owns the process and a shift team trained on a checklist rather than a habit.

The longer work closes the variation that lives in the people. Different crews prep differently. Night shift runs a different pre-change routine than day shift. Closing those gaps requires shift-level visibility on job change performance and coaching against a single agreed standard.

The systemised Job Change Tool Lean Glass runs is vendor-neutral. It's not tied to any equipment supplier's format and doesn't assume a specific SCADA platform. What it does assume is that you want a measured number at the end, not a report.

If you recognised your plant in the first two paragraphs, start by getting your last 10 job change timestamps in a room with your hot-end superintendent for two hours. That conversation alone usually finds 30 minutes. Reach out for a structured hot end audit if you want to go further.